Cursor is a code editor built for programming with AI. In versions 1.7.52 and below, manipulating internal settings may lead to RCE. Cursor detects path manipulation via forward slashes (./.cursor/./././././mcp.json etc.), and requires human approval to complete the operation. However, the same kind of manipulation using backslashes was not correctly detected, allowing an attacker who had already achieved prompt injection or some other level of control to overwrite sensitive editor files without approval on Windows machines. This issue is fixed in version 2.0.

Cursor is a code editor built for programming with AI. In versions below 1.3, an attacker can bypass the allow list in auto-run mode with a backtick (`) or $(cmd). If a user has swapped Cursor from its default settings (requiring approval for every terminal call) to an allowlist, an attacker can execute arbitrary command execution outside of the allowlist without user approval. An attacker can trigger this vulnerability if chained with indirect prompt injection. This is fixed in version 1.3.

Cursor is a code editor built for programming with AI. In versions 1.2.4 and below, attackers can achieve remote and persistent code execution by modifying an already trusted MCP configuration file inside a shared GitHub repository or editing the file locally on the target's machine. Once a collaborator accepts a harmless MCP, the attacker can silently swap it for a malicious command (e.g., calc.exe) without triggering any warning or re-prompt. If an attacker has write permissions on a user's active branches of a source repository that contains existing MCP servers the user has previously approved, or allows an attacker has arbitrary file-write locally, the attacker can achieve arbitrary code execution. This is fixed in version 1.3.

Cursor is a code editor built for programming with AI. Prior to 2.3, hen the Cursor Agent is running in Auto-Run Mode with Allowlist mode enabled, certain shell built-ins can still be executed without appearing in the allowlist and without requiring user approval. This allows an attacker via indirect or direct prompt injection to poison the shell environment by setting, modifying, or removing environment variables that influence trusted commands. This vulnerability is fixed in 2.3.

Cursor is a code editor built for programming with AI. In versions 1.7 and below, a vulnerability in the way Cursor CLI Agent protects its sensitive files (i.e. */.cursor/cli.json) allows attackers to modify the content of the files through prompt injection, thus achieving remote code execution. A prompt injection can lead to full RCE through modifying sensitive files on case-insensitive filesystems. This issue is fixed in a commit, 25b418f, but has yet to be released as of October 3, 2025.

Cursor is a code editor built for programming with AI. Versions 1.6 and below are vulnerable to Remote Code Execution (RCE) attacks through Visual Studio Code Workspaces. Workspaces allow users to open more than a single folder and save specific settings (pretty similar to .vscode/settings.json) for the folders / project. An untitled workspace is automatically created by VS Code (untitled.code-workspace), which contains all the folders and workspace settings from the user's current session, opening up an entire new attack vector if the user has a .code-workspace file in path (either untitled created automatically or a saved one). If an attacker is able to hijack the chat context of the victim (such as via a compromised MCP server), they can use prompt injection to make the Cursor Agent write into this file and modify the workspace. This leads to a bypass of CVE-2025-54130 which can lead to RCE by writing to the settings section. This issue is fixed in version 1.7.

Claude Code is an agentic coding tool. Prior to version 2.0.72, due to an error in command parsing, it was possible to bypass the Claude Code confirmation prompt to trigger execution of untrusted commands through the find command. Reliably exploiting this required the ability to add untrusted content into a Claude Code context window. This issue has been patched in version 2.0.72.

CI4MS is a CodeIgniter 4-based CMS skeleton that delivers a production-ready, modular architecture with RBAC authorization and theme support. Prior to version 0.28.5.0, an authenticated user with file editor permissions can achieve Remote Code Execution (RCE) by leveraging the file creation and save endpoints, an attacker can upload and execute arbitrary PHP code on the server. This issue has been patched in version 0.28.5.0.

A vulnerability has been found in composiohq composio up to 0.5.6 and classified as critical. Affected by this vulnerability is the function Calculator of the file python/composio/tools/local/mathematical/actions/calculator.py. The manipulation leads to code injection. The exploit has been disclosed to the public and may be used. The vendor was contacted early about this disclosure but did not respond in any way.

Skipper is an HTTP router and reverse proxy for service composition. The default skipper configuration before 0.23.0 was -lua-sources=inline,file. The problem starts if untrusted users can create lua filters, because of -lua-sources=inline , for example through a Kubernetes Ingress resource. The configuration inline allows these user to create a script that is able to read the filesystem accessible to the skipper process and if the user has access to read the logs, they an read skipper secrets. This vulnerability is fixed in 0.23.0.

Erugo is a self-hosted file-sharing platform. In versions up to and including 0.2.14, an authenticated low-privileged user can upload arbitrary files to any specified location due to insufficient validation of user‑supplied paths when creating shares. By specifying a writable path within the public web root, an attacker can upload and execute arbitrary code on the server, resulting in remote code execution (RCE). This vulnerability allows a low-privileged user to fully compromise the affected Erugo instance. Version 0.2.15 fixes the issue.

A vulnerability, which was classified as critical, was found in SourceCodester Simple Forum-Discussion System 1.0. Affected is an unknown function of the file /index.php. The manipulation of the argument page leads to path traversal. It is possible to launch the attack remotely. The exploit has been disclosed to the public and may be used.

AutoGPT is a platform that allows users to create, deploy, and manage continuous artificial intelligence agents that automate complex workflows. Prior to autogpt-platform-beta-v0.6.44, AutoGPT Platform's block execution endpoints (both main web API and external API) allow executing blocks by UUID without checking the `disabled` flag. Any authenticated user can execute the disabled `BlockInstallationBlock`, which writes arbitrary Python code to the server filesystem and executes it via `__import__()`, achieving Remote Code Execution. In default self-hosted deployments where Supabase signup is enabled, an attacker can self-register; if signup is disabled (e.g., hosted), the attacker needs an existing account. autogpt-platform-beta-v0.6.44 contains a fix.

Alist is a file list program that supports multiple storages, powered by Gin and Solidjs. Prior to version 3.57.0, the application contains path traversal vulnerability in multiple file operation handlers. An authenticated attacker can bypass directory-level authorisation by injecting traversal sequences into filename components, enabling unauthorised file removal, movement and copying across user boundaries within the same storage mount. This issue has been patched in version 3.57.0.

The SQL Expressions experimental feature of Grafana allows for the evaluation of `duckdb` queries containing user input. These queries are insufficiently sanitized before being passed to `duckdb`, leading to a command injection and local file inclusion vulnerability. Any user with the VIEWER or higher permission is capable of executing this attack. The `duckdb` binary must be present in Grafana's $PATH for this attack to function; by default, this binary is not installed in Grafana distributions.

The Frontend Dashboard plugin for WordPress is vulnerable to unauthorized code execution due to insufficient filtering on callable methods/functions via the ajax_request() function in all versions up to, and including, 2.2.4. This makes it possible for authenticated attackers, with subscriber-level access and above, to call arbitrary functions that can be leverage for privilege escalation by changing user's passwords.

Yoke is a Helm-inspired infrastructure-as-code (IaC) package deployer. In 0.19.0 and earlier, a vulnerability exists in the Air Traffic Controller (ATC) component of Yoke. It allows users with CR create/update permissions to execute arbitrary WASM code in the ATC controller context by injecting a malicious URL through the overrides.yoke.cd/flight annotation. The ATC controller downloads and executes the WASM module without proper URL validation, enabling attackers to create arbitrary Kubernetes resources or potentially escalate privileges to cluster-admin level.

PHP Everywhere <= 2.0.3 included functionality that allowed execution of PHP Code Snippets via WordPress shortcodes, which can be used by any authenticated user.

Improper limitation of a pathname to a restricted directory vulnerability in Samsung MagicINFO 9 Server version before 21.1050 allows attackers to write arbitrary file as system authority.

OpenList Frontend is a UI component for OpenList. Prior to 4.1.10, the application contains path traversal vulnerability in multiple file operation handlers in server/handles/fsmanage.go. Filename components in req.Names are directly concatenated with validated directories using stdpath.Join. This allows ".." sequences to bypass path restrictions, enabling users to access other users' files within the same storage mount and perform unauthorized actions such as deletion, renaming, or copying of files. An authenticated attacker can bypass directory-level authorisation by injecting traversal sequences into filename components, enabling unauthorised file removal and copying across user boundaries within the same storage mount. This vulnerability is fixed in 4.1.10.

An issue discovered in provectus kafka-ui 0.4.0 through 0.7.1 allows remote attackers to execute arbitrary code via the q parameter of /api/clusters/local/topics/{topic}/messages.

The Player Leaderboard plugin for WordPress is vulnerable to Local File Inclusion in all versions up to, and including, 1.0.2 via the 'player_leaderboard' shortcode. This is due to the plugin using an unsanitized user-supplied value from the shortcode's 'mode' attribute in a call to include() without proper path validation. This makes it possible for authenticated attackers, with Contributor-level access and above, to include and execute arbitrary PHP files on the server, allowing the execution of any PHP code in those files. This can be used to bypass access controls, obtain sensitive data, or achieve full remote code execution if combined with file upload capabilities.

Envoy Gateway is an open source project for managing Envoy Proxy as a standalone or Kubernetes-based application gateway. Prior to 1.5.7 and 1.6.2, EnvoyExtensionPolicy Lua scripts executed by Envoy proxy can be used to leak the proxy's credentials. These credentials can then be used to communicate with the control plane and gain access to all secrets that are used by Envoy proxy, e.g. TLS private keys and credentials used for downstream and upstream communication. This vulnerability is fixed in 1.5.7 and 1.6.2.

An authenticated remote code execution vulnerability in QStar Archive Solutions Release RELEASE_3-0 Build 7 Patch 0 allows attackers to arbitrarily execute commands.

Code injection via nginx.ingress.kubernetes.io/permanent-redirect annotation.

A CWE-22: Improper Limitation of a Pathname to a Restricted Directory ('Path Traversal') vulnerability exists that could allow a remote code execution when a file is saved. Affected Product: C-Bus Toolkit (V1.15.9 and prior), C-Gate Server (V2.11.7 and prior)

Remote code execution was discovered in Horde Groupware Webmail 5.2.22 and 5.2.17. Horde/Form/Type.php contains a vulnerable class that handles image upload in forms. When the Horde_Form_Type_image method onSubmit() is called on uploads, it invokes the functions getImage() and _getUpload(), which uses unsanitized user input as a path to save the image. The unsanitized POST parameter object[photo][img][file] is saved in the $upload[img][file] PHP variable, allowing an attacker to manipulate the $tmp_file passed to move_uploaded_file() to save the uploaded file. By setting the parameter to (for example) ../usr/share/horde/static/bd.php, one can write a PHP backdoor inside the web root. The static/ destination folder is a good candidate to drop the backdoor because it is always writable in Horde installations. (The unsanitized POST parameter went probably unnoticed because it's never submitted by the forms, which default to securely using a random path.)

Icinga Web 2 is an open source monitoring web interface, framework and command-line interface. Authenticated users, with access to the configuration, can create SSH resource files in unintended directories, leading to the execution of arbitrary code. This issue has been resolved in versions 2.8.6, 2.9.6 and 2.10 of Icinga Web 2. Users unable to upgrade should limit access to the Icinga Web 2 configuration.

Improper Control of Generation of Code ('Code Injection') vulnerability in Soft8Soft LLC Verge3D Publishing and E-Commerce.This issue affects Verge3D Publishing and E-Commerce: from n/a through 4.5.2.

The Custom Block Builder – Lazy Blocks plugin for WordPress is vulnerable to Remote Code Execution in all versions up to, and including, 4.2.0 via multiple functions in the 'LazyBlocks_Blocks' class. This makes it possible for authenticated attackers, with Contributor-level access and above, to execute code on the server.

A vulnerability was detected in abhiphile fermat-mcp up to 47f11def1cd37e45dd060f30cdce346cbdbd6f0a. This vulnerability affects the function eqn_chart of the file fmcp/mpl_mcp/core/eqn_chart.py. Performing a manipulation of the argument equations results in code injection. It is possible to initiate the attack remotely. The exploit is now public and may be used. This product is using a rolling release to provide continious delivery. Therefore, no version details for affected nor updated releases are available. The project was informed of the problem early through an issue report but has not responded yet.

Malicious code injection in Apache Ambari in prior to 2.7.8. Users are recommended to upgrade to version 2.7.8, which fixes this issue. Impact: A Cluster Operator can manipulate the request by adding a malicious code injection and gain a root over the cluster main host.

The DataEngine Xnode Server application in Zoho ManageEngine DataSecurity Plus prior to 6.0.1 does not validate the database schema name when handling a DR-SCHEMA-SYNC request. This allows an authenticated attacker to execute code in the context of the product by writing a JSP file to the webroot directory via directory traversal.

Ballcat Codegen provides the function of online editing code to generate templates. In versions prior to 1.0.0.beta.2, attackers can implement remote code execution through malicious code injection of the template engine. This happens because Velocity and freemarker templates are introduced but input verification is not done. The fault is rectified in version 1.0.0.beta.2.

Path traversal vulnerability exists in RakRak Document Plus Ver.3.2.0.0 to Ver.6.4.0.7 (excluding Ver.6.1.1.3a). If this vulnerability is exploited, arbitrary files on the server may be obtained or deleted by a user of the product with specific privileges.

Improper Control of Generation of Code ('Code Injection') vulnerability in Qode Interactive Qode Essential Addons.This issue affects Qode Essential Addons: from n/a through 1.5.2.

The CS Framework plugin for WordPress is vulnerable to arbitrary file deletion due to insufficient file path validation in the cs_widget_file_delete() function in all versions up to, and including, 6.9. This makes it possible for authenticated attackers, with Subscriber-level access and above, to delete arbitrary files on the server, which can easily lead to remote code execution when the right file is deleted (such as wp-config.php).

An issue discovered in OpenCart 4.0.0.0 to 4.0.2.3 allows authenticated backend users having common/security write privilege can write arbitrary untrusted data inside config.php and admin/config.php, resulting in remote code execution on the underlying server.

Combodo iTop is a web based IT Service Management tool. In versions prior to 2.7.6 and 3.0.0, users of the iTop user portal can send TWIG code to the server by forging specific http queries, and execute arbitrary code on the server using http server user privileges. This issue is fixed in versions 2.7.6 and 3.0.0. There are currently no known workarounds.

Improper Limitation of a Pathname to a Restricted Directory ('Path Traversal') vulnerability in QODE Interactive Qi Addons For Elementor allows PHP Local File Inclusion.This issue affects Qi Addons For Elementor: from n/a through 1.6.3.

The vulnerability allows a remote attacker to upload arbitrary files in all paths of the system under the context of the application OS user (“root”) via a crafted HTTP request. By abusing this vulnerability, it is possible to obtain remote code execution (RCE) with root privileges on the device.

PHP Everywhere <= 2.0.3 included functionality that allowed execution of PHP Code Snippets via a WordPress gutenberg block by any user able to edit posts.

pyLoad 0.5.0 is vulnerable to Unrestricted File Upload.

Flux is an open and extensible continuous delivery solution for Kubernetes. Path Traversal in the kustomize-controller via a malicious `kustomization.yaml` allows an attacker to expose sensitive data from the controller’s pod filesystem and possibly privilege escalation in multi-tenancy deployments. Workarounds include automated tooling in the user's CI/CD pipeline to validate `kustomization.yaml` files conform with specific policies. This vulnerability is fixed in kustomize-controller v0.24.0 and included in flux2 v0.29.0.

The serialize function used to compile MDX in next-mdx-remote is vulnerable to arbitrary code execution due to insufficient sanitization of MDX content. This vulnerability, CVE-2026-0969, is fixed in next-mdx-remote 6.0.0.

HotelDruid v3.0.3 was discovered to contain a remote code execution (RCE) vulnerability which is exploited via an attacker inserting a crafted payload into the name field under the Create New Room module.

Twig is an open source template language for PHP. When in a sandbox mode, the `arrow` parameter of the `sort` filter must be a closure to avoid attackers being able to run arbitrary PHP functions. In affected versions this constraint was not properly enforced and could lead to code injection of arbitrary PHP code. Patched versions now disallow calling non Closure in the `sort` filter as is the case for some other filters. Users are advised to upgrade.

This vulnerability could allow a remote attacker to execute remote commands with improper validation of parameters of certain API constructors. Remote attackers could use this vulnerability to execute malicious commands such as directory traversal.

OneDev is an all-in-one devops platform. In OneDev before version 4.0.3 there is a critical "zip slip" vulnerability. This issue may lead to arbitrary file write. The KubernetesResource REST endpoint untars user controlled data from the request body using TarUtils. TarUtils is a custom library method leveraging Apache Commons Compress. During the untar process, there are no checks in place to prevent an untarred file from traversing the file system and overriding an existing file. For a successful exploitation, the attacker requires a valid __JobToken__ which may not be possible to get without using any of the other reported vulnerabilities. But this should be considered a vulnerability in `io.onedev.commons.utils.TarUtils` since it lives in a different artifact and can affect other projects using it. This issue was addressed in 4.0.3 by validating paths in tar archive to only allow them to be in specified folder when extracted.

The Better Find and Replace – AI-Powered Suggestions plugin for WordPress is vulnerable to Limited Code Injection in all versions up to, and including, 1.7.7. This is due to insufficient input validation and restriction on the 'rtafar_ajax' function. This makes it possible for authenticated attackers, with Subscriber-level access and above, to call arbitrary plugin functions and execute code within those functions.